Abstract
Mutations in human VPS4A are associated with neurodevelopmental defects, including motor delays and defective muscle tone. VPS4A encodes a AAA-ATPase required for membrane scission, but how mutations in VPS4A lead to impaired control of motor function is not known. Here we identified a mutation in zebrafish vps4a, T248I, that affects sensorimotor transformation. Biochemical analyses indicate that the T248I mutation reduces the ATPase activity of Vps4a and disassembly of ESCRT filaments, which mediate membrane scission. Consistent with the role for Vps4a in exosome biogenesis, vps4aT248I larvae have enlarged endosomal compartments in the CNS and decreased numbers of circulating exosomes in brain ventricles. Resembling the central form of hypotonia in VPS4A patients, motor neurons and muscle cells are functional in mutant zebrafish. Both somatosensory and vestibular inputs robustly evoke tail and eye movements, respectively. In contrast, optomotor responses, vestibulospinal, and acoustic startle reflexes are absent or strongly impaired in vps4aT248I larvae, indicating a greater sensitivity of these circuits to the T248I mutation. ERG recordings revealed intensity-dependent deficits in the retina, and in vivo calcium imaging of the auditory pathway identified a moderate reduction in afferent neuron activity, partially accounting for the severe motor impairments in mutant larvae. Further investigation of central pathways in vps4aT248I mutants showed that activation of descending vestibulospinal and midbrain motor command neurons by sensory cues is strongly reduced. Our results suggest that defects in sensorimotor transformation underlie the profound yet selective effects on motor reflexes resulting from the loss of membrane scission mediated by Vps4a.
Significance Statement
Here we present a T248I mutation in vps4a, which causes sensorimotor defects in zebrafish larvae. Vps4a plays a key role in membrane scission. Spanning biochemical to systems level analyses, our study indicates that a reduction in Vps4a enzymatic activity leads to abnormalities in membrane scission-dependent processes such as endosomal protein trafficking and exosome biogenesis, resulting in pronounced deficits in sensorimotor transformation of visual, auditory, and vestibular cues. We suggest that the mechanisms underlying this type of dysfunction in zebrafish may also contribute to the condition seen in human patients with de novo mutations in the human VPS4A ortholog.
Introduction
Neurodevelopmental disorders are a collection of conditions that affect central nervous system growth and development and can have a wide variety of associated clinical signs, such as impaired cognition, difficulties in communication, and motor defects. Many neurodevelopmental disorders are the result of genetic mutations and often target molecular pathways that impact protein synthesis, transcriptional or epigenetic regulation, or synaptic signaling (Lewis and Kroll, 2018; Cardoso et al., 2019; Parenti et al., 2020). An early feature that is common to several neurodevelopmental defects is central hypotonia (low muscle tone); however, it remains unclear how the underlying biochemical and sensorimotor mechanisms can result in this feature (Dan, 2022).
De novo heterozygous missense mutations in human vacuolar protein sorting-associated protein 4A (VPS4A) have been associated with neurodevelopmental disorders characterized by pronounced delays in development, motor skills and speech, and severe intellectual disabilities and visual dysfunction (Rodger et al., 2020). Symptoms arising from problems with muscle tone such as hypotonia and dystonia are also thought to be central in nature (Ganguly et al., 2021). In addition, mutations in VPS4A are associated with congenital dyserythropoietic anemia (Seu et al., 2020; Lunati et al., 2021). VPS4A encodes a type I AAA-ATPase that was originally identified in yeast. In multicellular organisms, a paralog, VPS4B, is also expressed, and both genes are highly conserved from plants to humans (Babst et al., 1997; Scheuring et al., 2001; Strausberg et al., 2002; Beyer et al., 2003). VPS4A binds to the endosomal sorting complexes required for transport (ESCRT) fission machinery to promote dissociation of the subunits via ATP hydrolysis after membrane scission occurs. Membrane scission is a key step in several cellular processes including cytokinesis, neurite severing, plasma membrane repair, and multivesicular body formation (Babst et al., 2002; Monroe et al., 2014; Alonso et al., 2016; Ott et al., 2018). Multivesicular bodies form within the endocytic pathway when late endosomes bud inward and membrane scission occurs, generating small intraluminal vesicles (Piper and Katzmann, 2007; Alonso et al., 2016; Bebelman et al., 2020; Kalluri and LeBleu, 2020). During further processing, multivesicular bodies can subsequently either fuse to the plasma membrane to release the intraluminal vesicles as exosomes into the extracellular matrix or fuse with lysosomes to promote degradation of its contents (Dobrowolski and De Robertis, 2012; Meldolesi, 2018). Both fates of intraluminal vesicles are key for maintaining cell homeostasis and signal transduction (Cocucci et al., 2009; Palmulli and van Niel, 2018). In animal studies, it has been previously shown that RNAi knockdown of vps4 in fruit fly larvae results in defective pruning of neurites and knock-out of Vps4a in mice results in lethality at preweaning stages (Loncle et al., 2015; Cacheiro et al., 2020). In line with the known function of VPS4A, fibroblasts from VPS4A human patients display defects in endosomal morphology and cell division (Rodger et al., 2020). How these defects lead to motor system impairment in probands is not fully understood.
Here, we identified a missense mutation in zebrafish vps4a, T248I, which was generated in a small-scale ENU mutagenesis screen for mutants that display auditory/vestibular defects (Haffter et al., 1996). We show that the zebrafish vps4a mutants display endosomal abnormalities and reduced numbers of exosomes in the CNS. Both defects are consistent with the induction of stress response genes in mid- and hindbrain regions of vps4aT248I mutants. Using in vivo calcium imaging, we detected strongly reduced activity in the motor circuits associated with movements evoked by sensory inputs. These findings support a model where the reduced enzymatic activity of Vps4a results in a disruption in late endosome function and exosome release, consequently reducing cellular homeostasis within neural circuits mediating sensorimotor transformation.
Materials and Methods
Zebrafish care and use
We maintained zebrafish lines for the vps4aT248I allele in the Top Long Fin (TLF) wild-type (WT) or mitfab692 nacre backgrounds. The vps4aT248I Tg(α-tubulin-gal4;MP-gal4;cmlc2-GFP;UAS-GCaMP7a) line was generated by crossing into a mitfab692 Tg(α-tubulin-gal4;MP-gal4;cmlc2-GFP;UAS-GCaMP7a) fish line (a gift from Philippe Mourrain). The MP-gal4 transgene was present in most fish, marking the muscle cells as well. Breeding stocks were housed at 28.5°C and animal husbandry followed standard zebrafish methods for laboratory utilization (Westerfield, 2000), as approved and overseen by the Institutional Animal Care and Use Committees at both Oregon Health and Sciences University and Stanford University. The experiments used zebrafish larvae 3–9 days postfertilization (dpf) before sexual differentiation occurs. Embryos and larvae were raised in E3 medium (in mM: 0.33 CaCl2, 0.17 KCl, 0.33 MgSO4, and 5 NaCl) incubated at 28.5°C. When appropriate, larvae were anesthetized in E3 with 0.03% 3-amino benzoic acid ethylester (MESAB, Western Chemical) to minimize pain and distress.
Genotyping methods
For the vps4aT248I allele, we genotyped fish and larvae using kompetitive allele-specific PCR (KASP) genotyping (LGC Biosearch Technologies) per the manufacturer’s instructions.
Identification of raumschiff mutant allele
Positional cloning was performed using simple sequence length polymorphism (SSLP) markers mapped to the zebrafish genome (Knapik et al., 1998). Five dpf and older zebrafish larvae were sorted by phenotype.
For whole-genome sequencing, 5 dpf zebrafish larvae were sorted into separate pools of 20 homozygous mutants and 20 heterozygous and homozygous wild-type siblings. Genomic DNA was collected using a DNeasy Blood & Tissue Kit (Qiagen), and samples were processed by Azenta Life Sciences (formerly GENEWIZ) for library preparation and Illumina HiSeq sequencing. The resulting .gz files were uploaded to SNPTrack (http://genetics.bwh.harvard.edu/snptrack/; Leshchiner et al., 2012) to identify the mutation.
CRISPR injections
Single-guide RNAs against vps4a and vps4b were designed using the CHOPCHOP website (https://chopchop.cbu.uib.no/; Labun et al., 2019) and synthesized (Alt-R CRISPR-Cas9 System, IDT). Three guides were chosen based on if they targeted the ATPase domain or additional exons common to all isoforms, low number of off target sites, and efficiency. Multiguide CRISPR injections were performed as described previously (Kroll et al., 2021). Briefly, crRNA and tracrRNA guides were combined in equal molar amounts in Duplex Buffer (IDT), and this RNA Duplex mix was combined with 1 μg/μl Cas9-NLS protein (PNA Bio). Then, 1 nl of RNA guide-CAS9 mixture was injected into single-cell stage TLF wild-type embryos. Embryos were checked daily for viability and auditory/vestibular phenotypes scored on 5 dpf.
Protein expression and purification
Saccharomyces cerevisiae Vps4 (wild-type) and mutant Vps4(T254I) were cloned into the pET28a vector with an N-terminal His6 tag. The two proteins were expressed in Escherichia coli BL21 (DE3) strains and induced with 1 mM IPTG at 20°C overnight. Sonication was used to lyse cells, and the lysates were centrifuged. The supernatant was loaded into Ni–NTA beads (GE Healthcare), and the eluates were incubated with 5 U ml−1 apyrase (Sigma-Aldrich) at 4°C overnight. All proteins were further purified by SEC using a Superdex 200 10/300 GL column (GE Healthcare) with 20 mM Tis-HCl at pH 8.0, 150 mM NaCl buffer. The purity and quality of all the proteins were tested by SDS–PAGE and Nano-drop (Implen).
S. cerevisiae Snf7 mutant (R52E) and Vps24 were cloned into the pET28a vector with an N-terminal His6 tag. The two proteins were expressed in E. coli BL21 (DE3) strains and induced with 1 mM IPTG at 20°C overnight. Sonication was used to lyse cells and the lysates were centrifuged. The supernatant was loaded to a Ni–NTA beads (GE Healthcare) and the elute was further purified by SEC using a Superdex 200 10/300 GL column (GE Healthcare) with 20 mM Tis-HCl at pH 8.0, 150 mM NaCl buffer. The purity and quality of all the proteins were tested by SDS–PAGE and Nano-drop (Implen).
S. cerevisiae Vps2 were cloned into the pET28a vector with an N-terminal MBP-His6 -TEV-tag. We expressed the resulting Vps2 construct in E. coli BL21 (DE3) at 20°C for 16 h with 1 mM IPTG. We purified the protein using Ni–NTA resin (GE Healthcare). We mixed the eluted fractions from the His column with TEV (tobacco etch virus) protease for 3 h at 4°C. We then loaded the cleaved samples onto a 5 ml HiTrap Q-sepharose FF column (GE Healthcare) and eluted the material with a gradient of 0–100% buffer (20 mM Tris-HCl, pH 8.0, 500 mM NaCl) in 50 column volumes.
Sedimentation analysis of filament disassembly
Yeast ESCRT-III filaments were assembled by mixing Snd7(R52E), Vps24, and Vps2 in a molar ratio of 2:1:1 at room temperature for 1 h. Then, the formed filaments were centrifuged at 50,000 rpm for 30 min in a TLA-100 ultracentrifuge rotor (Beckman Coulter). The pellet was resuspended in the buffer containing 20 mM Tris-HCl at pH 8.0, 150 mM NaCl. The pellet (P) and supernatant (S) were analyzed by SDS–PAGE.
For sedimentation analysis of ESCRT-III filament disassembly, filaments were mixed with 5 μM Vps4 or Vps4(T254I) in 20 mM Tris-HCl at pH 8.0, 150 mM NaCl, 2 mM MgCl2 buffer, separately. Reactions were started by adding ATP to a final concentration of 5 mM. After incubation at room temperature for 10 min, reactions were stopped by adding EDTA to a final concentration of 50 mM. Samples were then subjected to ultracentrifugation and analyzed via SDS–PAGE and negative staining.
ATPase activity assay
ATPase activities were determined for wild-type and mutant Vps4 proteins using the QuantiChrom ATPase/GTPase Assay Kit (BioAssay Systems) at room temperature. Briefly, 5 μM Vps4 or mutant Vps4(T254I) protein was added to ATPase assay buffer and 4 mM ATP for 30 min and then stopped using reagent from the kit. The solution was incubated at room temperature for 30 min, and immediately the absorbance at 620 nm was detected using a plate reader (EnSpire, PerkinElmer). The released phosphate was calculated based on the absorbance standard curve established using KH2PO4 standards.
Auditory-evoked behavioral response
Auditory-evoked behavioral response (AEBR) was tested as described previously (Erickson et al., 2017; Smith et al., 2020) using the Zebrabox system (ViewPoint Life Sciences). Five dpf wild-type siblings or vps4aT248I mutant larvae were confined to wells of a 96-well plate in 200 μl E3 per well, in the dark. Larvae were subjected to 100 millisecond, 1 kHz pure tones at 157 decibel sound pressure (dB SPL, relative to 1mPA) every 2 s for 1 min. An infrared camera recorded larval movements and ZebraLab tracking software documented pixel changes, which represent movement over time for each larva. To distinguish between evoked responses and spontaneous background movement, movements within 1 s before a stimulus were excluded from a given trial. For each larva, the positive responses were calculated as a percentage out of total possible response within a minute, and the best of three trials was used for analysis.
Vestibulospinal reflex
The device and method used to test the vestibulospinal reflex (VSR) was done as previously described (P. Sun et al., 2018; Gao and Nicolson, 2021). Briefly, 5 dpf larvae were embedded dorsal-up in 2% prewarmed low melt agarose in E3 media without anesthetic on a mounting chamber. The agarose around the trunk and pectoral fins was carefully removed and replaced with 20 μl E3 media, and the mounting chamber was placed in the device. A motor rocks the platform from −75 to +75° at 0.53 Hz, and an infrared camera recorded video of the reflex at 30 fps (1,280 frames).
Modified ZebraZoom software (Mirat et al., 2013) tracks the movements of the trunk at regular intervals and calculates the real tail angle over time between the x-axis and the axis formed by the tip of the tail and the center of the swimming bladder.
Vestibular-induced eye movement
The vestibular-induced eye movement (VIEM) experiment was done similarly as described previously using a modified device (Mo et al., 2010; P. Sun et al., 2018). Then, 5 dpf larvae were embedded on a small coverslip in prewarmed 2% low melt agarose in E3 buffer, without anesthetic. Agarose surrounding the eyes was gently removed and replaced with ∼10 μl of E3 media to permit free eye moment. The mounted larvae were positioned vertically, head down, between an infrared camera, and an infrared LED array on a programmable rocking platform. Larvae kept in darkness during the experiment. The platform rocked (−45 to 45°) at 0.25 Hz frequency for 60 s while larval eye movements were video recorded in infrared.
Software-based video analyses quantified the amplitudes of angular eye movements as a function of their frequency of repetition. Larvae with normal VIEM show peak movement amplitude at 0.25 Hz, in synchrony with platform rotation. Baseline spontaneous eye movement, taken as the average amplitude across all quantified frequencies, was subtracted to determine the VIEM amplitude at 0.25 Hz. We calculated the normalized amplitude of the eye ratio for each larva and used the two-tailed unpaired t test with Welch's correction to compare mutants with their corresponding siblings.
Touch startle reflex
Touch startle was performed by mounting 7 dpf zebrafish larvae similarly to the VSR mounting method. A micromanipulator holding a borosilicate glass capillary tube (Sutter Instrument, #BF150-86-10) pulled into a needle was used to touch the head of the larvae. Startle movements were recorded by a Hamamatsu C11440 ORCA-Flash 2.8 camera at 100 fps. Quantification of C-bends was performed as described in Budick and O’Malley (2000).
Optomotor reflex
To test for the optomotor reflex (OMR), we adapted the assay from Neuhauss et al., (1999). A set of four tracks (each lane measuring at 12 × 0.6 cm) and comb that fit between the lanes was constructed using 3D printing. The lanes were placed in a 15 cm petri dish (VWR) and filled partially with E3 buffer. The comb was inserted to block the starting end of the lanes, and 5−6 dpf larvae were placed in the starting area of the lanes. The petri dish was placed over a screen that plays a video of 0.5 cm stripes. To record the movement of the larva along the lanes, a camera was mounted above the petri dish. To perform the experiment, the video of stripes was played, and the comb was removed, allowing larva to swim against the movement of stripes to the end of the lane. To quantify this reflex, screenshots of the video were taken of the larva in the starting region of the lane prior to comb removal and then 30 s after the comb was removed. The distance between the location of each larva prior to comb removal and 30 s after was measured in ImageJ in millimeters.
Spontaneous swimming
Then, 6 dpf mutant and wild-type sibling larvae were placed in 6-well plates filled with E3 buffer. Spontaneous swimming was recorded by a camera mounted above the dish and recorded for 1 min. Traces of spontaneous swimming were manually recorded using ImageJ, and the distance traveled was quantified using the Skeletonize plugin to record running average lengths.
Electroretinogram
Electroretinogram (ERG) recordings were performed on 5–6 dpf mutant and wild-type sibling larvae, as described in (Makhankov et al., 2004; Niklaus et al., 2017). The b-wave amplitudes were plotted and analyzed using Dunn's multiple comparisons.
Immunohistochemistry
Larvae were fixed with 4% paraformaldehyde (PFA), dehydrated with methanol, rehydrated using a graded series of PBSDT (1× PBS with 0.25% Tween 20 and 1% DMSO): methanol and then treated with 150 mM Tris-HCl for antigen retrieval. Samples were then washed and permeabilized with ice-cold 100% acetone, followed by more washes. Samples were blocked for 2 h at room temperature with 6% goat serum and 3% BSA and then incubated overnight with 1:1,000 anti-acetylated tubulin rabbit primary antibodies (Thermo Fisher Scientific). Samples were then incubated overnight with 1:1,000 anti-rabbit Alexa 488 (Invitrogen) secondary antibodies, protected from light.
LysoTracker
Five dpf PTU-treated larvae were sorted and placed in mutant–sibling pairs in a 2 ml Eppendorf tube. LysoTracker Red DND-99 (Invitrogen) was diluted to 1:100 to examine brains or 1:200 to examine inner ear hair cells. Larvae were incubated in LysoTracker for 40 min at 28.5°C with the tube left open but protected from light. The larvae were then washed with E3 buffer for 10 min and immediately imaged.
Hindbrain LysoTracker uptake was quantified by selecting a single Z-stack slice of the hindbrain region and thresholding it to highlight particles in the size range of endosomal compartments (size, 0.02–2 μm). The integrated density and area of particles from two equally sized regions of interest from hindbrain sections and a single region of neuropil were measured from each slice using the Analyze Particles function of ImageJ. The integrated density measurements were averaged and normalized to the integrated density of the neuropil.
CD63-pHluorin injections and analysis
pUbi-CD63-pHluorin was a gift from Frederik Verweij and was generated as described in Verweij et al. (2019). Nacre-background vps4aT248I mutant and sibling embryos were injected at the one-cell stage with 100 ng/μl pUbi-CD63-pHluorin. Damaged embryos were sorted out, and only larvae with CD63-pHluorin expression on muscle cell membranes were selected for live imaging on 3–4 dpf. Time-lapse videos were recorded at 50 fps for 2 min at 63× with a 2× optical zoom. For analysis, three frames were combined, and the mean pixel intensity was measured from regions of the ventricular intracranial space, surrounding tissue as background, and muscle cells. The mean pixel intensity of ventricular intracranial space was normalized to background and then divided by muscle cell pixel intensity.
Bulk tissue RNA-seq
Total RNA was isolated from 5 dpf wild-type and vps4a mutant larvae (n = 15–20 each genotype) using TRIzol reagent (Thermo Fisher Scientific, 15596026) according to the manufacturer's protocol. RNA-seq library construction and single-end sequencing were performed by the OHSU Massively Parallel Sequencing Shared Resource (RRID SCR_009984, Oregon Health & Science University). Sequencing reads were mapped to the Danio rerio GRCz10 genome using Tophat v2.1.1 (Kim et al., 2013) from within an OSHU instance of the Galaxy sequence analysis platform (Afgan et al., 2016). The resulting BAM files were imported to SeqMonk v1.42.0 (www.bioinformatics.babraham.ac.uk/projects/seqmonk/) for data visualization and statistical analysis using the Intensity Difference filter for unreplicated datasets. Genes with an adjusted p value of <0.05 were considered differentially expressed.
Hybridization chain reaction
For hybridization chain reaction (HCR), custom probes for vps4a, atf3, jun, gap43, opn1sw1, and the corresponding fluorescent B-amplifier hairpins were ordered from Molecular Instruments (https://www.molecularinstruments.com/). Probes were chosen based on their log2(fold change) and number of reads from bulk RNA sequencing analysis. Probes were assigned B-amplifiers with the intention of doing multiplex experiments. Probe sequences are considered confidential and proprietary information of Molecular Instruments based on their Terms and Conditions, so further inquiries may be directed to their technical team.
PTU-treated larvae were sorted at 5 dpf and fixed overnight in 4% PFA at 4°C with rocking. After fixation, larvae were washed in PBST (1× PBS with 0.1% Tween 20), dehydrated with a methanol:PBS series (25, 50, and 75% methanol in PBS), and then stored at least overnight at −20°C. Larvae were subsequently rehydrated with a reverse methanol:PBST series, washed several times with PBST, permeabilized for 15 min with 20 μg/ml proteinase K, postfixed with 4% PFA, and then washed again. Probe detection and amplification with fluorescent hairpins was performed as recommended by the MI protocol for whole-mount zebrafish larva with extended prehybridization and preamplification times (4 h each) and probe concentrations adjusted based on the reads of each probe (2 nM for atf3 and jun, 5 nM for gap43, 10 nM for vps4a).
Calcium imaging
Larvae at 3 dpf vps4aT248I Tg(α-tubulin-gal4:UAS-GCaMP7a) were sorted for strong GCaMP expression in the brain. Five dpf GCaMP-positive larvae were mounted in 2% low melt agarose on a custom depression slide fitted with a mini-speaker (DAEX-9-45M Skinny Mini Exciter 9 mm 1 W 4 Ohm). Using an AIYIMA A07 amplifier, tone stimuli were played through the mini-speaker using a custom script in MATLAB (R2020b, MathWorks). Then, 600 Hz tones with a 0.6 s duration were generated using a 0.2 V driver voltage resulting in 125 dB stimuli (calibrated using a waterproof mini-microphone (WP-23502-P16, Knowles Electronics). With respect to g forces, 0.2 V driver voltage yielded 10 g stimuli [calibrated using a fiber laser Doppler vibrometer (LDV); Model: OFV501, Polytec]. This stimulus frequency and intensity yielded reliable calcium transients in all regions in wild-type specimens, including the dimmest region of the statoacoustic ganglion. For each individual fish, five tones with 10 s intervals were delivered during imaging. For spinal cord imaging, we used five 450 Hz tones at 125 dB for 0.1 s. Time-lapse images and fluorescence values were recorded using LAS X software (ver 3.7.2.22383). To account for variability in GCaMP7a expression between larvae, all fluorescence values were normalized by subtracting background measurements except in the case of the statoacoustic ganglion, where low signal-to-background noise masked the percentage change in fluorescence. Traces were generated by averaging the normalized or raw fluorescence value (F) for each time point across all larvae and plotting this value over time. To calculate percentage change in fluorescence (ΔFt/F0%), 2–3 frames containing image artifacts (due to the tone stimulation shaking the slide) were identified for each stimulus, and fluorescence values corresponding to these frames were discarded. The F0 measurement was selected from the frame prior to tone stimulation, based on the time-lapse video and presence of artifact frames, and the Ft measurement was selected from a frame with the highest signal within 1 s after tone stimulation. The change in GCaMP7a fluorescence was calculated using ΔFt/F0, where ΔFt = Ft − F0. ΔFt/F0 was multiplied by 100 to convert into a percentage to generate (ΔFt/F0%).
The average percentage change in fluorescence evoked by five tones was then calculated for each larva, and differences between wild-type siblings and vps4a mutants were analyzed using unpaired t tests.
Retrograde labeling
Retrograde labeling was adapted from previously published methods (Gahtan et al., 2002; Wehner et al., 2017). Larvae were anesthetized with 0.3% MESAB prior to spinal injection of 25 mg/ml 10,000 MW tetramethylrhodamine dextran (10,000 MW, D1866, Thermo Fisher Scientific) in PBS. Custom glass injection needles were prepared from borosilicate glass capillaries (TW100F-4, World Precision Instruments) using a microneedle puller (Sutter Instrument, Model P-87) using heat cycle 517, pulling cycle 75, velocity 75, and time 90, and then the tip was broken using forceps. Larvae were mounted laterally in 2% low melt agarose for injection and incubated in E3 media overnight to recover before imaging.
Microscopy and image processing
Images for LysoTracker, CD63-pHluorin, and HCR were taken using an upright LSM700 laser scanning confocal microscope with Zeiss Zen software (Carl Zeiss). For live imaging of larva, they were anesthetized in 0.03% MESAB in E3 media and mounted in 2% prewarmed low melt agarose. Images were collected using 488, 555, and 647 nm lasers, depending on the requirements of the individual experiment (acetylated tubulin and CD63-pHluorin used 488 nm, LysoTracker used 555 nm, HCR used up to all three depending on the fluorescent hairpins used). We used a Leica HC PL Fluotar 10×/0.3 lens for acetylated tubulin, HCR, and LysoTracker images, a Plan-Apochromat 40×/1.0DIC lens for LysoTracker quantification images and an Achroplan 63×/0.95 W water immersion lens for higher magnification HCR images. Laser power and gain were determined independently for each fluorophore and kept consistent within each experiment. The FIJI build of ImageJ version 2.1.0/1.53t was used to process all raw image files, quantify abnormal cells, and measure head and brain sizes of larvae. For calcium imaging, scans were taken using a Leica SP8 Confocal Microscope equipped with a 20× (NA 1.0) objective, using LAS X software. Scans were collected at 600 Hz unidirectional scan speed with pixel/voxel size 1.083 µm, at 0.293 s per frame, using 488 and 555 nm lasers.
Experimental design and statistical analysis
The number of zebrafish larvae used is denoted in the figure legends. G*Power software was used to ensure there were sufficient samples tested. All statistical analyses were performed using GraphPad Prism software (version 9). The data was tested for normal distribution using Shapiro–Wilk test. Data is presented as mean ± SD, and significance between mutant and wild-type sibling groups was determined using unpaired t test unless otherwise noted. The image in Figure 1C was adapted from S. Sun et al. (2017) using Mol* (Sehnal et al., 2021) and RCSB PDB (PDB ID:5XMI) software.
The T248I mutation in vps4a reduces ATPase activity and ESCRT-III filament disassembly. A, vps4a is located on Chromosome 25 in a location corresponding to where classical mapping methods have identified the location of the gene responsible for the raumschiff phenotypes. B, Bar diagram of Vps4a protein showing the location of the T248I mutation. C, Atomic model structure of the Vps4E233Q hexamer, with T248 marked in magenta (Image adapted from the RCSB PDB ID:5XMI, using Mol*). D, Mutant Vps4 (T254I) yeast protein has decreased ATP activity compared with wild-type Vps4 protein. Unpaired t test; p < 0.0001; nWT = 4; nvps4(T254I) = 4; t = 47.92; df = 6. E, Sedimentation analysis of Vps4 and Vps4(T254I)-mediated disassembly of ESCRT-III filaments. Pellet (P) and supernatant (S) fractions were analyzed by SDS–PAGE. Marker (M). F, Negative staining EM micrograph of sedimentation analysis of ESCRT-III filament disassembly.
Results
A missense mutation in vps4a reduces ATPase activity and ESCRT-III filament disassembly
raumschiff mutants were generated from an ENU mutagenesis screen for genes that affect larval auditory and vestibular function (Haffter et al., 1996). Mutant larvae fail to respond to auditory cues and maintain an upright posture. The mutation is recessive and fully penetrant at 5 dpf with lethality occurring ∼10 dpf. To identify the causative gene of the raumschiff phenotype, we used whole-genome sequencing and identified a threonine to isoleucine missense mutation in vps4a using SNPTrack (Leshchiner et al., 2012). The critical interval for raumschiff was also mapped to a region ∼700 kb (or 1 cm) south of the polymorphic marker z5669, placing the locus at 34.33 Mb on Chromosome 25. vps4a is located at 34.93 Mb on Chromosome 25, which is within the predicted region for the causative gene. No other polymorphisms were identified by SNPTrack in this region (Fig. 1A).
Vps4a is a hexameric type I AAA-ATPase member of the ESCRT-III complex, whose role is to promote the dissociation of the ESCRT-III subunits via ATP hydrolysis (Babst et al., 1997; Davies et al., 2010; Monroe et al., 2014; McCullough et al., 2018). The threonine to isoleucine mutation (T248I) we identified is present in the AAA-ATPase domain in a region that is highly conserved between species (Fig. 1B). To determine whether the T248I mutation was comparable with a nonsense allele of vps4a, we edited the vps4a locus using multiple CRISPR/Cas9 guides (Wu et al., 2018; Kroll et al., 2021). In our experiments, multiguide injections resulted in a G0 phenotype in 16.9% of injected larvae that was indistinguishable from the auditory/vestibular phenotype observed in vps4aT248I mutants (400 embryos were injected at < 1 h post fertilization; of the larvae that survived to 5 dpf, 56/331 larvae showed absence of acoustic startle reflexes and abnormal posture). These results suggest that the T248I mutation is comparable with a null allele of vps4a. In addition, the vps4b paralog is also ubiquitously expressed (Yang et al., 2016); therefore, we generated vps4a/vps4b G0 double mutants using CRISPR-based editing to test whether vps4b could compensate for the loss of vps4a during development. A previous study using morpholinos against vps4b reported a tooth phenotype analogous to human dentin dysplasia I (Yang et al., 2016). We observed that embryos injected with multiple CRISPR guides against vps4b alone, or vps4a and vps4b together, developed normally until 3 dpf. The lack of embryonic lethality is presumably due to the presence of maternal RNA during early developmental stages. However, in contrast to the morpholino study, vps4b and vps4a/4b multiguide injected larvae exhibited progressive necrosis and by 5 dpf, there was complete lethality (558 embryos were injected with vps4b guides; 468 larvae survived to 1 dpf, and 2 survived to 5 dpf. 298 embryos were injected with vps4a/4b guides; 236 larvae survived to 1 dpf, and 1 survived to 5 dpf). Our results suggest that vps4b is required for global cellular homeostasis whereas vps4a selectively affects the nervous system. A similar effect was reported for the Nsfa and Nsfb AAA-ATPases that mediate vesicle fusion; mutations in zebrafish nsfa largely affect the nervous system, while the loss of nsfb functions leads to general necrosis at 4 dpf (Woods et al., 2006; Hanovice et al., 2015; Gao et al., 2023).
Based on the Cryo-EM structure of the yeast Vps4 AAA-ATPase (S. Sun et al., 2017), threonine 248 is located at the interface between Vps4a hexamer subunits (Fig. 1C). To assess the effect of the isoleucine substitution on ATPase activity, we generated an equivalent mutation, T254I, in the AAA-ATPase domain of yeast Vps4. When tested for ATPase activity in vitro, Vps4T254I had severely reduced enzymatic activity compared with wild-type Vps4a, although some activity remained (p < 0.0001; nWT = 4; nvps4(T254I) = 4; t = 47.92; df = 6; Fig. 1D).
To determine whether the residual function of the T254I protein could mediate filament disassembly, we incubated Vps4T254I with the ESCRT-III filaments formed by Vps2, Vps24, and Snf7 in vitro and used sedimentation to assess disassembly. In our assay, disassembled ESCRT-III filament proteins are present in the supernatant whereas assembled filaments sediment to the pellet fraction (S. Sun et al., 2017). We observed that wild-type Vps4a completely disassembled the ESCRT-III filaments, with all proteins found in the supernatant (Fig. 1E). In contrast, mutant Vps4T254I only partially disassembled the ESCRT filaments (Fig. 1E). Using negative stain electron microscopy (EM), assembled ESCRT-III filaments appear as rings or have corkscrew structures (Fig. 1F, top left panel). When incubated with wild-type Vps4, filament structures were absent, indicating full disassembly (Fig. 1F, top right panel). In the presence of Vps4T254I, assembled filaments were still detectable albeit there were fewer filaments compared with samples where no Vps4 protein was added (Fig. 1F, bottom panels). These results demonstrate that the T245I mutation at the interface between Vps4a subunits reduces but does not fully abolish the ATPase activity of the protein, and this reduction in activity is reflected in the incomplete disassociation of ESCRT-III filaments.
The vps4aT248I mutation results in enlarged endosomal membrane compartments and a reduction of circulating exosomes in the CNS
Using RNA-FISH, we found that the expression of vps4a transcripts is ubiquitous at 5 dpf, with higher levels expressed in the CNS and retina (Fig. 2A). In light of the higher expression of vps4a in the CNS and the neurological disorder in human patients, we focused our attention on the larval brain. As stated previously, Vps4a mediates the formation of multivesicular bodies from late endosomes. To determine if the mutation affects this function in the nervous system, we performed in vivo imaging of LysoTracker, a vital dye that accumulates in acidic membrane compartments such as low pH endosomal compartments. As shown in Figure 2B, endocytic compartments in vps4aT248I mutants labeled more brightly with LysoTracker compared with wild-type siblings. The average size and pixel density of acidic membrane compartments in the hindbrain were quantified and we found that vps4aT248I mutants have significantly larger acidic membrane compartments (p = 0.0015; nWT = 7; nvps4a(T248I) = 10; t = 3.873; df = 15; Fig. 2C, left panel) that take up more dye (p = 0.0020; nWT = 7; nvps4a(T248I) = 10; t = 3.744; df = 15; Fig. 2C, right panel). Together, these results imply that the vps4aT248I mutation has a detrimental effect on processing of endocytic compartments in the CNS.
vps4aT248I mutants have enlarged endosomal compartments and release fewer exosomes in the CNS. A, Confocal Z-projections of a 5 dpf wild-type zebrafish larva labeled with HCR RNA-FISH probes for vps4a. fb, forebrain; ot, optic tectum; tpz, tectal proliferation zone, cb, cerebellum; hb, hindbrain. B, LysoTracker staining of 5 dpf wild-type siblings and vps4aT248I mutants show an increased staining in the mutant. C, Acidic membrane compartment size in the hindbrain of vps4aT248I mutants (N = 10) is significantly larger than those in wild-type siblings (N = 7); unpaired t test; p = 0.0015; t = 3.873; df = 15. LysoTracker vital dye taken up by vps4aT248I mutants (N = 10), measured as normalized pixel integrated density, is significantly higher compared with wild-type siblings (N = 7); unpaired t test; p = 0.0020; t = 3.744; df = 15. D, Schematic of injecting one-cell stage nacre-background embryos with pUbi:CD63-pHluorin. Healthy larvae were sorted on 1 dpf, and larvae expressing CD63-pHluorin in muscle cells was sorted on 3 dpf, for imaging on 4 dpf. E, Decreased CD63-pHluorin expression in 3 dpf vps4aT248I mutant CNS blood vessels. F, Dorsal view of 4 dpf zebrafish larva showing the midbrain ventricle. G, Decreased CD63-pHluorin expression in vps4aT248I mutant brain ventricles, but trunk muscles express comparable levels to wild-type siblings. H, Pixel intensity of muscles cells is correlated with normalized pixel intensity of CD63-pHluorin signal in the midbrain ventricle in wild-type siblings (N = 16). I, Quantification of CD63-pHluorin fluorescence in wild-type siblings (N = 16) and vps4aT248I mutants (N = 8); Mann–Whitney test; p = 0.0382 (exact).
Given the known role of Vps4a in multivesicular body formation, we examined whether the mutation affects the number of exosomes that are released upon fusion of multivesicular bodies to the plasma membrane. To detect circulating exosomes in live fish, we used CD63-pHluorin, a pH-sensitive GFP reporter that is quenched in intraluminal vesicles in low pH multivesicular bodies and fluoresces when these vesicles are released into a higher pH extracellular milieu as exosomes (Verweij et al., 2018). We microinjected a plasmid for ubiquitous expression of CD63-pHluorin into single-cell stage embryos and selected for larvae that transiently expressed the CD63-pHluorin based on signal in the muscle cells, which is readily detectable by 3 dpf (Fig. 2D). The plasma membrane of muscle cells was commonly labeled in our injected fish as was previously reported (Verweij et al., 2018) and served as a positive control for expression. We detected a strong reduction in CD63-pHluorin signal in the CNS blood vessels on 3 dpf (Fig. 2E), as well as in the midbrain ventricle of 4 dpf vps4aT248I mutants (Fig. 2F,G). We quantified the CD63-pHluorin signal in the midbrain ventricle and normalized to background signal from the surrounding CNS tissue. The relationship between CD63-pHluorin signal in the midbrain ventricle is moderately correlated to signal in the muscle cells in wild-type siblings but not in mutants (Fig. 2H). We then calculated a ratio of the normalized pixel intensity of the midbrain ventricle CD63-pHluorin signal to the muscle cell CD63-pHlourin signal and found that the ventricle/muscle cell intensity ratio is significantly decreased in vps4aT248I mutants compared with that in wild-type siblings [p = 0.0382 (exact); nWT = 16; nvps4a(T248I) = 8; Mann–Whitney test; Fig. 2I]. This result indicates that there is a decrease in the number of exosomes present in the brain ventricle of mutants, supporting the hypothesis that biogenesis of these extracellular vesicles is reduced by the vps4aT248I mutation.
Transcriptional changes in the vps4aT248I mutant
Considering the broad changes in endosomal morphology and the reduction in circulating exosomes in the CNS of vps4aT248I mutants, we next examined the gross morphology of the brain. Using whole-mount immunohistochemistry, we labeled the axonal fibers of mutants and siblings with acetylated tubulin antibody to visualize axon morphology. No obvious changes in gross morphology between wild-type siblings and vps4aT248I mutants were apparent at 5 dpf (Fig. 3A). The lack of obvious developmental defects is likely due to the expression of vps4b or maternal vps4a mRNA, presumably allowing normal development to proceed during embryonic development. However, we noted the presence of pyknotic nuclei (Fig. 3B), which are associated with cell death (Hou et al., 2016). We quantified the number of these abnormal cells and found that vps4aT248I mutants have significantly more abnormal cells in the CNS compared with wild-type siblings and that cell death increases during development [5 dpf, p < 0.0001 (exact), nWT = 12, nvps4a(T248I) = 14, Mann–Whitney test; 7 dpf, p < 0.0001 (exact), nWT = 15, nvps4a(T248I) = 18, Mann–Whitney test; 9 dpf, p < 0.0001 (exact), nWT = 9, nvps4a(T248I) = 13, Mann–Whitney test]. We observed that the size of the head (9 dpf; p < 0.0001; nWT = 15; nvps4(T254I) = 18; t = 6.933; df = 31) and brain (9 dpf; p < 0.0001; nWT = 16; nvps4(T254I) = 18; t = 4.754; df = 32) were decreased in 9 dpf vps4aT248I mutants compared with that in wild-type siblings, but not at 7 dpf (head, p = 0.3979, nWT = 12, nvps4(T254I) = 15, t = 0.8602, df = 25; brain, p = 0.8127, nWT = 12, nvps4(T254I) = 15, t = 0.2395, df = 25; Fig. 3C). The decrease in head and brain size in the vps4aT248I mutant is consistent with progressive cell death.
vps4aT248I mutants upregulate stress response and regeneration genes in the CNS. A, Axons labeled with acetylated tubulin antibody staining of vps4aT248I mutants show no gross morphological defects. Fb, forebrain; ot, optic tectum; mb, midbrain; cb, cerebellum; hb, hindbrain. B, vps4aT248I mutants have an increase of morphologically abnormal, pyknotic cells in the CNS. Two examples are highlighted with black dashed circles. The number of abnormal cells in the vps4aT248I mutants was quantified and found to be significantly higher than in wild-type siblings at both 5 and 7 dpf; Mann–Whitney test; 5 dpf, p < 0.0001 (exact), nWT = 12, nvps4a(T248I) = 14; 7 dpf, p < 0.0001 (exact), nWT = 15, nvps4a(T248I) = 18; 9 dpf, p < 0.0001 (exact), nWT = 9, nvps4a(T248I) = 13. C, vps4aT248I mutants have a smaller head (unpaired t test, 7 dpf, p = 0.3979, t = 0.8602, df = 25, nWT = 12, nvps4a(T248I) = 15; 9 dpf, p < 0.0001, t = 6.933, df = 31, nWT = 15, nvps4a(T248I) = 18) and brain (unpaired t test, 7 dpf, p = 0.8127, t = 0.2395, df = 25, nWT = 12, nvps4a(T248I) = 15; 9 dpf, p = 0.0105, t = 2.722, df = 31, nWT = 15, nvps4a(T248I) = 18) at 9 dpf. The ROI for the head measurement is outlined in pink and the ROI for the brain measurement is outlined in yellow. D, Heat map showing Log2 fold change gene expression in vps4aT248I mutants relative to control WT siblings. E, Maximum intensity projections of atf3 and jun and gap43 and jun mRNA expression in the hindbrain of 5 dpf wild-type sibling and vps4aT248I mutants.
As exosomes carry cargo containing microRNA and DNA and have been implicated in the regulation of DNA transcription (Kalluri and LeBleu, 2020), we looked for transcriptional changes in the vps4aT248I mutant by performing a differential gene expression analysis on a bulk RNA-seq dataset that was initially used to find the mutation. We then used HCR RNA-FISH to validate a subset of the top log2 fold changes in gene expression (Fig. 3D). We found that the immediate early response genes atf3 and jun and the axonal regeneration gene gap43 were dramatically upregulated in mid- and hindbrain regions in the vps4aT248I larvae (Fig. 3E). These striking changes in gene expression are consistent with a failure in intercellular communication by exosomes and/or endosomal processing within neurons.
Impaired vision in vps4aT248I mutants
The clinical profile of human patients with missense mutations in VPS4A includes visual deficits with a range of symptoms such as retinal dystrophy and cataract formation (Rodger et al., 2020; Seu et al., 2020). Although we did not observe expanded melanophores in the vps4aT248I mutant, which are indicative of blindness in larvae, we did find intensity-dependent changes in visual function and gene expression in the retina. The optomotor response (OMR) in zebrafish is a reflex that can be evoked by a visual stimulus such as moving stripes, which causes the fish to swim in the direction of perceived motion, resulting in stabilization relative to the environment (Kist and Portugues, 2019). Compared with wild-type siblings, vps4aT248I mutant larvae have a greatly reduced OMR, suggesting they are blind [p < 0.0001 (exact); nWT = 12; nvps4a(T248I) = 10; Mann–Whitney test; Fig. 4A]. To investigate this defect further, we recorded ERGs to assess the light-induced changes in the electrical potential of the whole retina. The b-wave amplitude generated from an ERG can provide insight into the function of the outer retina (Niklaus et al., 2017). A range of light intensities (from the dimmest at log −4, to the brightest at log 0) were tested, and we found that the b-wave amplitude in vps4aT248I mutants is significantly decreased compared with that in wild-type siblings (log 0, p < 0.0001, nWT = 23, nvps4a(T248I) = 34; log −1, p < 0.0001, nWT = 23, nvps4a(T248I) = 33; log −2, p < 0.0001, nWT = 23, nvps4a(T248I) = 34; log −3, p = 0.0014, nWT = 23, nvps4a(T248I) = 34, Kruskal–Wallis test). However, at the lowest light level, there was no significant difference (log −4, p = 0.1989, nWT = 23, nvps4a(T248I) = 34, Kruskal–Wallis test; Fig. 4B). With respect to changes in gene expression, we found that jun and gap43 are upregulated in the retina of vps4aT248I mutants, specifically in the photoreceptor and retinal ganglion layers (Fig. 4C). We also detected reduced opn1sw1 expression in the photoreceptors of mutants. Together, these analyses indicate that the vps4aT248I phenotype includes visual defects originating in the retina. Notably, the b-wave amplitude is not significantly different at the lowest intensity of light, and the decrease in b-wave amplitude at higher intensities is variable among individual mutants. Thus, the partial reduction in retinal activity does not fully explain the absence of the OMR.
Impaired vision in vps4aT248I mutants. A, 5 dpf vps4aT248I mutants (N = 10) have a severely reduced optomotor response compared with wild-type siblings (N = 12); Mann–Whitney test; p < 0.0001 (exact). B, Decreased b-wave amplitude in ERGs of 5–6 dpf vps4aT248I mutants (N = 34) in comparison with wild-type siblings (N = 23) occurs at brighter light intensities (log −4 to log 0); Kruskal–Wallis test (log 0, p < 0.0001; log −1, p < 0.0001; log −2, p < 0.0001; log −3, p = 0.0014; log −4, p = 0.1989). C, HCR RNA-FISH of gap43, jun, and opn1sw1 in the eye of vps4aT248I mutants compared with wild-type siblings. Pr, photoreceptors; Rg, retinal ganglion cells, lens.
Selective vestibular defects in vps4aT248I mutants
At the free-swimming stage of development (5 dpf), vps4aT248I larvae do not maintain an upright posture, swim sideways or upside down, and do not inflate their swim bladder, which is classically associated with functional deficits in vestibular hair cells (Nicolson, 2017). To further investigate this phenotype in vps4aT248I mutant larvae, we tested vestibular function by assessing VIEMs (Mo et al., 2010). Larvae with loss of hair-cell function mutations have reduced eye movements in response to head rotation (Nicolson, 2017); however, we did not observe a significant difference between the vps4aT248I mutants and wild-type siblings [p = 0.4632 (exact); nWT = 18; nvps4a(T248I) = 16; Mann–Whitney test; Fig. 5A]. As mutant larvae have pronounced postural defects, we also tested the VSR, which assesses vestibular-induced motor function in the trunk (Gao and Nicolson, 2021). We compared the maximum tail angle between wild-type siblings and mutants and found that the vps4aT248I mutants had a significantly reduced tail angle [p = 0.0012 (exact); nWT = 7; nvps4a(T248I) = 10; Mann–Whitney test; Fig. 5B]. The normal VIEM along with the reduced VSR result suggests that vps4aT248I mutants have selective defects in vestibular pathways.
vps4aT248I mutants have selective defects in vestibular circuitry. A, 5 dpf vps4aT248I mutants (N = 16) do not have a significantly different VIEM compared with wild-type siblings (N = 18); Mann–Whitney test; p = 0.4632 (exact). B, 5 dpf vps4aT248I mutants (N = 10) have a reduced VSR compared with wild-type siblings (N = 7); Mann–Whitney test; p = 0.0012 (exact). C, Simplified schematic showing the flow of information evoked by head rotation from the inner ear to hindbrain nuclei, vestibuloocular neurons, and motor neurons. UHC, utricular hair cells; CN VIII, VIIIth nerve; SVN, superior vestibular nucleus; VS, vestibulospinal cells; TAN, tangential neurons; M, Mauthner cell; CN IV, trochlear nucleus; CN III, oculomotor nucleus; OM, oculomotor muscles. D, 5 dpf larva showing basal GCaMP7a expression and backfill injected rhodamine dextran marking the Mauthner cell (M). White dotted lines indicate the presumptive TAN, SVN, and VS regions, and a solid yellow line designates a dextran filled VS neuron. Other landmarks include fibers running parallel and perpendicular fibers to the inner ear cavity (yellow arrows), a lateral cleft between mid- and hindbrain lobes (white arrowhead) and portion of the occipital arch cartilage (yellow asterisk). D’, Pre- and poststimulus GCaMP7a levels in vestibular region in a representative wild-type sibling and vps4aT248I mutant. The TAN, SVN, and VS ROIs are outlined by dotted lines. E, Average trace of normalized fluorescence (F) for the TAN, SVN, and VS of wild-type siblings and vps4aT248I mutants during auditory stimulation at 600 Hz. F, Percentage change in fluorescence of the presumptive TAN, SVN, and VS of wild-type siblings and vps4aT248I mutants during auditory stimulation at 600 Hz. The VS in vps4aT248I mutants (N = 11) have a reduced change in fluorescence compared with wild-type siblings (N = 15); unpaired t test; p = 0.0337; t = 2.252; df = 24. Single VS neurons in vps4aT248I mutants (N = 9 cells from 4 larva) also have a reduced change in fluorescence compared with wild-type siblings (N = 26 cells from 14 larva); unpaired t test; p = <0.0001; t = 6.195; df = 33. There is no significant difference between mutants and wild-type siblings in the TAN (unpaired t test; p = 0.1405; t = 1.524; df = 24) or SVN (unpaired t test; p = 0.6224; t = 0.4988; df = 24).
In zebrafish, acceleration of the head is detected by utricular hair cells (Tanimoto et al., 2009; Mo et al., 2010; P. Sun et al., 2018), which then transmit signals to the utricular afferents of the VIIIth cranial nerve (Fig. 5C). The VIIIth nerve innervates the superior vestibular nucleus (SVN), vestibulospinal nucleus (VS), and tangential nucleus (TAN), as well as Mauthner cells in the hindbrain (Bianco et al., 2012; Jia and Bagnall, 2022; Liu et al., 2022; Hamling et al., 2023b). Signals from the SVN and TAN are relayed by the oculomotor and trochlear nuclei to oculomotor muscles to produce eye movements associated with the VIEM (Fig. 5C; Bianco et al., 2012; Liu et al., 2022; Sugioka et al., 2023). VS neurons extend axons into the spinal cord and presumably directly innervate motor neurons in the trunk to produce tail movements (Liu et al., 2022; Sugioka et al., 2023). To assess the function of the VSR circuit, we used a pan-neuronal GCaMP7a calcium indicator to visualize neuronal activity (Muto et al., 2013). Recently it was demonstrated that utricular hair cells in zebrafish larvae respond to a wide range of frequencies (2–2,000 Hz), analogous to the responses of vestibular hair cells to bone-conducted vibration in mammals (Tanimoto et al., 2022; P. Sun et al., 2024). To stimulate the vestibular circuitry mediating the VIEM and VSR, we delivered pulses of 600 Hz at 10 g using a mini-speaker attached to the glass slide on which larvae were mounted for calcium imaging. Reconstruction of larval vestibular nuclei in the hindbrain using electron microscopy has shown that the lateral dendrite of the Mauthner cell is adjacent to the VS nucleus, which is located between the SVN and TAN (Liu et al., 2022). To visualize this dendrite, we performed spinal backfill injections of rhodamine dextran prior to calcium imaging, enabling us to use the Mauthner cell as a landmark for the approximate locations of the SVN, VS, and TAN (Fig. 5D). In some cases, VS neurons were labeled as well (Fig. 5D). We selected regions of interest indicated in Figure 5D based on several criteria: (1) the position of the Mauthner cell dendrite (indicated with “M”), (2) the presence of fiber networks that ran parallel and perpendicular to the inner ear cavity (Fig. 5D, yellow arrows), (3) the curvature of the dorsal otic cavity, (4) the presence of a stereotypical lateral cleft between mid- and hindbrain lobes (white arrowhead), and (5) the presence of a small portion of the occipital arch cartilage near the posterior edge of the otic cavity (yellow asterisk). In addition, our regions were conservatively defined to avoid any overlap between nuclei. Upon tone stimulation, neurons in all three presumptive nuclei responded with an increase in fluorescence (Fig. 5D’). Nevertheless, the average raw fluorescence trace for the VS region has a lower peak amplitude in the vps4aT248I mutant compared with that in wild-type siblings (Fig. 5E, right). Analysis of the group data from individual fish indicates that the percentage change in normalized fluorescence in the VS region is significantly reduced in vps4aT248I mutants (p = 0.0337; nWT = 11; nvps4a(T248I) = 15; t = 2.252; df = 24; Fig. 5F). To avoid any potential contribution by Mauthner cell dendrites to the signal in the VS region, we also assessed calcium transients in single, backfilled somas of VS neurons (Fig. 5D). Single-cell analysis revealed a pronounced decrease in activity in VS neurons in vps4aT248I mutants (p = <0.0001; nWT cells = 26; nvps4a(T248I) cells = 9; t = 6.195; df = 33; Fig. 5F, right).
Serving as an internal control and predicted by our behavioral analysis of the VIEM, we did not observe a significant difference in the raw fluorescence trace or percentage change in normalized fluorescence in the presumptive TAN or SVN between vps4aT248I mutants and wild-type siblings [TAN p = 0.6224; SVN p = 0.1405; nWT = 11; nvps4a(T248I) = 15; t = 0.4988 (SVN); df = 24 (SVN); t = 1.524 (TAN); df = 24 (TAN); Fig. 5E,F]. Overall, our calcium imaging data confirm that the VIEM circuit is intact and suggest that reduced activity in the VS neurons, which presumably activate motor neurons, contributes in part to the reduction in tail movements in response to vestibular cues in the vps4aT248I mutants.
Moderate neural deficit in the ascending auditory pathway in vps4aT248I mutants
In zebrafish, auditory signals excite saccular hair cells, which are innervated by the afferent neurons of the statoacoustic ganglion (or VIIIth nerve). If the auditory cues are sufficiently loud enough to startle, the signal is transmitted to the giant escape fiber neuron known as the Mauthner cell, which directly activates motor neurons (Kohashi and Oda, 2008; Sillar, 2009; López-Schier, 2019). Less intense sounds activate hindbrain neurons within the medial octavolateralis nuclei region, which then communicate with the torus semicircularis for further auditory processing (Echteler, 1984; Tanimoto et al., 2009; Inoue et al., 2013; Vanwalleghem et al., 2017; Fig. 6A). vps4aT248I mutant larvae are largely unresponsive to tapping on the petri dish, suggesting that hearing is profoundly impaired. We assessed the severity of this defect by quantifying the AEBR, which is a startle response to loud, pure tone stimuli (1 kHz, 157 dB). vps4aT248I mutant larvae responded to loud, pure tones significantly less compared with wild-type siblings, indicating that mutants have pronounced auditory startle deficits (p < 0.0001 (exact); nWT = 15; nvps4a(T248I) = 20; Mann–Whitney test; Fig. 6B). Next, we used the pan-neuronal GCaMP7a calcium indicator to examine neuronal activity in the saccular hair cells and statoacoustic ganglion in response to the lower intensity 110 dB 600 Hz stimulus (Fig. 6C,D). In saccular hair cells, we observed robust responses in average raw fluorescence traces (Fig. 6E) and comparable calcium transients in wild-type or mutant individual fish (p = 0.4162; nWT = 20; nvps4a(T248I) = 13; t = 0.8228; df = 35; Fig. 6F). Tone-evoked responses were also seen in the statoacoustic ganglion, however, they were partially reduced by 40% (p = 0.0067; nWT = 20; nvps4a(T248I) = 12; t = 2.911; df = 30; Fig. 6E,F). Thus, in terms of peripheral cell types, there is a moderate reduction in the neural component of the pathway.
Activation of the ascending auditory pathway in vps4aT248I mutants. A, Simplified schematic showing the flow of information from an auditory stimulus to motor neurons or to the torus semicircularis. S, saccular otolith; SHC, saccular hair cells; CN VIII, VIIIth nerve; MON, medial octavolateralis nucleus; M, Mauthner cell; TS, torus semicircularis. B, 5 dpf vps4aT248I mutants (N = 20) have a greatly reduced AEBR compared with wild-type siblings (N = 15); Mann–Whitney test; p < 0.0001 (exact). C, Basal levels of GCaMP7a expression in the inner ear of 5 dpf larvae, focusing on the saccular hair cells (dotted line) and statoacoustic ganglion (SAG; dashed line). D, Pre- and poststimulus GCaMP7a levels in saccular hair cells and SAG in a representative wild-type sibling and vps4aT248I mutant larvae. E, Average traces of normalized fluorescence (F) for the saccular hair cells and raw fluorescence (F) for the SAG of wild-type siblings and vps4aT248I mutants during auditory stimulation at 600 Hz. F, Percentage change in fluorescence of the saccular hair cells and SAG of wild-type siblings and vps4aT248I mutants during auditory stimulation at 600 Hz. There is no significant difference between mutants (N = 13) and wild-type siblings (N = 20) in the saccular hair cells (unpaired t test; p = 0.4162; t = 0.8228; df = 35). However, there is a partial decrease in percentage change fluorescence in the SAG (unpaired t test; p = 0.0067; t = 2.911; df = 30). G, Representative image of basal levels of GCaMP7a expression in 5 dpf zebrafish focusing on the TS (torus semicircularis, thick dashed line) and anatomical landmarks: MLF, medial longitudinal fascicles; LLF, lateral longitudinal fascicles (white arrowheads); OT, ventral edge of optic tectum (thin dashed line); S, saccule. Midbrain clefts are marked with an asterisk. H, Pre- and poststimulus GCaMP7a levels in the TS in representative wild-type sibling and vps4aT248I mutant larvae. I, Average trace of normalized fluorescence (F) for the TS of wild-type siblings and vps4aT248I mutants during auditory stimulation at 600 Hz. J, There is no significant difference in percentage change fluorescence between vps4aT248I mutants (N = 13) and wild-type siblings (N = 11) in the TS; unpaired t test; p = 0.9672; t = 0.04157; df = 22.
Next, we determined the effect of the partial reduction in afferent neuron activity on the central components of the ascending pathway. Auditory circuitry within the hindbrain for nonstartle responses is less well defined in larvae, therefore we focused on a midbrain region of the ascending auditory pathway and examined neuronal activity in the torus semicircularis, which is akin to the inferior colliculus in mammals and key for auditory processing (McCormick, 1999; Edds-Walton and Fay, 2008; Vanwalleghem et al., 2017; Privat et al., 2019). This midbrain structure is located beneath the optic tectum at the level of the semicircular canals and saccule of the inner ear and the medial longitudinal fasciculus (MLF; Fig. 6G). We imaged optical sections of the dorsal region of the TS below the optic tectum using the following anatomical landmarks: (1) the presence of both the MLF and lateral longitudinal fascicles, (2) the ventral edge of the optic tectum that partially surrounds the TS, (3) the cleft between the mid-hindbrain, and (4) the saccule along with the anterior and posterior semicircular canals (Fig. 6G). We observed a visible increase in GCaMP7a reporter activity in the presumptive torus semicircularis of both wild-type siblings and vps4aT248I mutants after pure tone stimulation (Fig. 6H). No significant differences were detected in the average raw fluorescence traces or percentage change in normalized fluorescence between wild-type siblings and vps4aT248I mutants for the torus semicircularis (p = 0.9672; nWT = 11; nvps4a(T248I) = 13; t = 0.04157; df = 22; Fig. 6I,J). Together our data suggest that, despite the moderate reduction in afferent neuron activity, the higher order TS region of the auditory pathway is still activated in vps4aT248I mutants by tone stimulation.
Defective sensorimotor transformation in vps4aT248I mutants
Our experiments indicate that vps4aT248I mutants have a greatly reduced acoustic startle response, yet auditory cues can still activate the ascending auditory pathway. Notably, mutant larvae are not paralyzed; however, they are largely inactive at the free-swimming stage. Therefore, we assessed whether vps4aT248I mutants have a general motor system defect that prevents motor responses to sensory cues.
To test whether startle reflexes are globally affected, we measured the ability of vps4aT248I mutants to produce a deep C-bend of the trunk in response to contact of a stiff probe with the dorsal region of the head. We observed that both wild-type siblings and vps4aT248I mutants displayed robust C-bend motions of their trunks in response to touch (Fig. 7A,B; p = 0.1079; nWT = 10; nvps4a(T248I) = 11; t = 1.687; df = 19). This result suggests that touch receptors, along with Mauthner cells and other vital components of motor output such as spinal cord neurons, motor neurons, and muscle cells are unaffected. As further evidence that the motor system is intact, we also imaged spontaneous swimming bouts in the mutants. Despite pronounced vestibular dysfunction, vps4aT248I mutants exhibited spontaneous movements, albeit some larvae traveled shorter distances and in tighter trajectories (Fig. 7C; p = 0.5063; nWT = 6; nvps4a(T248I = 5; t = 0.6921; df = 9). Furthermore, we did not observe any changes in the gross morphology of axonal fibers in superficial layers of the skin, the hindbrain, and anterior trunk regions in vps4aT248I mutants (Fig. 7D). These data point to a deficit within central circuitry in vps4aT248I mutants as downstream motor components of the startle reflex are functional.
vps4aT248I mutants display defects in sensorimotor transformation. A, 7 dpf vps4aT248I mutant larvae have an intact startle reflex in response to aversive touch stimuli. B, 7 dpf vps4aT248I mutants (N = 11) do not have a significantly different C-bend angle compared with wild-type siblings (N = 10); unpaired t test; p = 0.1079; t = 1.687; df = 19. C, Traces of spontaneous swimming from 6 dpf wild-type siblings and mutants over the course of 1 min. vps4aT248I larvae (N = 5) swim shorter distances compared with wild-type siblings (N = 6); unpaired t test; p = 0.5063; t = 0.6921; df = 9) D, Acetylated tubulin antibody staining of vps4aT248I mutants shows no gross morphological defects in hindbrain/spinal cord or somatosensory axons. E, Simplified schematic showing the flow of information from a tone stimulus from the inner ear to motor neurons via Mauthner cell and nMLF pathways. S, saccular otolith; SHC, saccular hair cells; CN VIII, VIIIth nerve; M, Mauthner cell; CRN, cranial relay neurons; nMLF, nucleus of the medial longitudinal fasciculus; TAN, tangential neurons; UHC, utricular hair cells. F, 5 dpf larva showing basal GCaMP7a expression and backfill injected rhodamine dextran marking single cells within the nMLF. The nMLF region is outlined with a dotted line. G, Pre- and poststimulus GCaMP7a levels in the nMLF in a representative wild-type sibling and vps4aT248I mutant. H, Average trace of normalized fluorescence (F) for the nMLF of wild-type siblings and vps4aT248I mutants during tone stimulation at 600 Hz. I, vps4aT248I mutants (N = 11) have a significantly reduced percentage change fluorescence in the nMLF compared with wild-type siblings (N = 14); unpaired t test; p < 0.0001; t = 6.937; df = 23. Single backfilled neurons of the nMLF in vps4aT248I mutants (N = 18 cells from 6 larvae) also have a reduced change in percentage fluorescence compared with wild-type siblings (N = 20 cells from 9 larvae); unpaired t test; p < 0.0001; t = 8.018; df = 36. J, Pre- and poststimulus GCaMP7a levels in the spinal cord motor neurons of representative wild-type sibling and vps4aT248I mutant. K, Average trace of normalized fluorescence (F) for the spinal cord motor neurons of wild-type siblings and vps4aT248I mutants during tone stimulation at 450 Hz. L, vps4aT248I mutants (N = 9) have a significantly reduced percentage change fluorescence in spinal cord neurons compared with wild-type siblings (N = 11), Mann–Whitney test; p < 0.0001 (exact).
As described above, auditory signals from the statoacoustic ganglion are directly transmitted to the Mauthner cell, which innervates trunk motor neurons to generate an escape reflex (Fig. 7E). However, the 110 dB intensity of our tone stimulus did not result in detectable activation of the Mauthner cell soma, which is consistent with a previous imaging study (Marsden and Granato, 2015). Instead, we observed a reliable response in the midbrain nucleus of the medial longitudinal fasciculus (nMLF; Fig. 7F–I). The nMLF is a vital component of the descending motor system, serving as a command center for forward locomotion (Sankrithi and O’Malley, 2010; Koyama et al., 2011; Severi et al., 2014; Thiele et al., 2014; Wang and McLean, 2014; Migault et al., 2018; Shimazaki et al., 2019; Xu et al., 2021). The nMLF contains subpopulations of vGlut1 and vGlut2 positive neurons, which control both fast escape-swimming behaviors as well as slow forward locomotion (Severi et al., 2014; Berg et al., 2023). Neurons in the nMLF were recently shown to respond to hindbrain cranial relay neurons (CRN; Fig. 7E), which can be activated by aversive auditory stimulation of Mauthner cells, leading to smooth swimming after an initial C-bend response (Xu et al., 2021). Although the auditory circuitry to the nMLF activated by nonstartle stimuli has not been characterized, the nMLF has been implicated in fine postural control via inputs from the TAN of the ascending vestibuloocular pathway (Sugioka et al., 2023). Notably, activation of neurons in the TAN in vps4aT248I mutants is comparable with wild-type responses.
Considering that the nMLF is central to motor function in response to sensory cues, we compared GCaMP7a signals in this cluster of neurons between wild-type siblings and vps4aT248I mutants in response to the 600 Hz stimulus. These midbrain clusters can be identified based on their prominent axonal tracts (MLF) that descend near the midline, as well as the landmarks used to locate the TS (Fig. 7F). We found that the peak amplitudes of calcium transients in the average raw fluorescence trace in the mutant were markedly decreased (Fig. 7G). In addition, the percentage change in normalized fluorescence in the nMLF was significantly reduced in all individual vps4aT248I mutants tested [p < 0.0001 (exact); nWT = 14; nvps4a(T248I) = 11; t = 6.937; df = 23; Fig. 7I, left). In larvae where single neurons in the nMLF were labeled from spinal backfills with rhodamine dextran, we also observed reduced activity in the cell body [p < 0.0001 (exact); nWT = 20; nvps4a(T248I) = 18; t = 8.018; df = 36; Fig. 7I, right).
We next investigated the neuronal activity of the spinal cord in response to tone stimulation using the pan-neuronal GCaMP7a reporter (Fig. 7J). The average raw fluorescence trace of the vps4aT248I mutants show lower peaks in response to tone stimulation (Fig. 7K). In the examples shown in Figure 7J, a second reporter (GCaMP7a expressed under a muscle cell promoter; Leung et al., 2019) in the trunk muscles shows activation in the wild-type larva but none in the mutant. Overall, the percentage change in normalized fluorescence is strongly reduced in vps4aT248I mutants [p < 0.0001 (exact); nWT = 11; nvps4a(T248I) = 9; Mann–Whitney test; Fig. 7L]. Together, our results indicate that tone stimulation fails to trigger normal responses in the nMLF descending motor pathway and downstream trunk motor neurons in vps4aT248I mutants.
Discussion
Our study identifies a role for vps4a in sensorimotor transformation during larval development in zebrafish. The T248I mutation in vps4a, which was isolated in a forward genetic screen for balance and hearing deficits, profoundly affects enzymatic function, leading to defects in ESCRT filament disassembly and membrane scission-dependent processes such as exosome biogenesis. In vps4aT248I larvae, we observe an increase in the expression of jun and atf3 stress response genes and a progressive loss in cellular homeostasis in the developing brain and retina. Although identified as a deafness mutant, the function of the inner ear in vps4aT248I larvae is unaffected. Combined behavioral analyses and in vivo imaging of circuit activity demonstrate that the ascending auditory pathway and descending motor pathways are functional in vps4aT248I mutants, however, mutants do not respond to acoustic stimuli. These data suggest that loss of vps4a function induces early onset of deficits in the transformation of sensory signals, particularly auditory, vestibular, and visual input into motor responses.
Defective disassembly of ESCRT filaments and decreased exosome biogenesis in zebrafish vps4a mutants
Our biochemical analyses indicate that the substitution of the corresponding threonine residue with isoleucine in the ATPase domain of yeast Vps4 strongly reduces ATPase activity and ESCRT-III filament disassembly. In vivo imaging of the CNS revealed enlargement of acidic compartments in larval mutants, which is consistent with previous studies in yeast vps4 mutants and human VPS4A patient fibroblasts (Bishop and Woodman, 2000; Frankel et al., 2017; González et al., 2017; Willén et al., 2017). The increased size of endosomal compartments is presumably due to the disruption of endosomal processing and when membrane fails to invaginate and form intraluminal vesicles via the ESCRT-mediated scission process. Intraluminal vesicles can be released from nearly all cell types as exosomes and contain a variety of signaling molecules, such as proteins, lipids, and nucleic acids (Colombo et al., 2014; Y. Zhang et al., 2019; Gurunathan et al., 2021; Waqas et al., 2022). Exosomes have been implicated in a several diseases such as cancers and neurodegenerative diseases due to the vital role they play in intercellular communication (Lim and Lee, 2017; Bebelman et al., 2020; Dilsiz, 2020; Gomes et al., 2020; Duarte-Silva et al., 2022; Han et al., 2023). A previous study on erythroid cells, which found that transferrin receptor trafficking, a process dependent on proper endosomal sorting and exosome secretion, is disrupted in human patients with de novo VPS4A mutations (Seu et al., 2020). Here we show direct evidence for reduced exosome biogenesis using a CD63-pHluorin marker of circulating exosomes. Both in blood vessels and brain ventricles, vps4aT248I mutants show a marked decrease in CD63-pHlourin signal. Disruptions to intercellular signaling through exosomes can manifest as transcriptional changes (Dobrowolski and De Robertis, 2012; Budnik et al., 2016; Gurung et al., 2021). We found that several genes associated with transcription (atf3 and jun) or regeneration (gap43) were highly upregulated in the CNS of vps4aT248I mutants. atf3 and jun are known stress response genes, whereas upregulation of gap43 may be a compensatory response to disruption of axonal processes. Although exosomes mediate cell-to-cell signaling and are important for cellular homeostasis, it is unclear whether the transcriptional changes in vps4aT248I mutants are due to (1) reduced exosome signaling, (2) disruptions in endosomal trafficking, or (3) a combination of both defects. Alternatively, other disruptions to membrane scission in the CNS may be involved as well. The expression of stress response genes in the CNS in zebrafish mutants during larval development is consistent with the microcephaly seen in human patients. While we did not observe changes in gross morphology of the CNS at early larval stages (5 dpf), cell death in the brain was evident and increased during larval development. As a likely consequence of cellular degeneration, we noted smaller heads and brains in vps4a mutants at 9 dpf. We speculate that more pronounced defects would be evident in older animals, however, the mutation is lethal at 10 dpf and we limited our analyses to the initial stages of larval development during which we could identify early onset defects.
Sensory and motor deficits
Patients with VPS4A mutations are reported to have sensory deficits relating to vision including retinal dystrophy and cataracts (Rodger et al., 2020). With respect to visual function in vps4aT248I mutants, larvae fail to swim in response to moving visual cues, and outer retinal responses to light are significantly reduced at higher intensities, signifying defects in vision that align with transcriptional changes in the retina. The reduction in outer retinal function was variable among mutants, whereas the optomotor response was severely reduced in all vps4aT248I mutants, suggesting that loss of photoreceptor function provides only a partial explanation of the absence of vision-associated motor reflexes. Notably, human patients are reported to have difficulties in fixing and following a moving object, which may be due to poor visual acuity or a combination of poor vision and oculomotor pathway dysfunction.
Loss of central control of muscle tone, ataxia, and motor delays are common traits found in patients with neurodevelopmental disorders, including VPS4A patients (Rodger et al., 2020; Seu et al., 2020). vps4aT248I larvae share some of these features, such as lack of control over posture and sensorimotor deficits. VSRs, which result in corrective movements of the trunk, are greatly diminished in vps4a mutants. However, not all vestibular function is lost; vps4aT248I larvae still produce eye movements in response to rotation of the head. Consistent with the behavioral data, presumptive neurons of the TAN and SVN, which are part of vestibuloocular pathway (Bianco et al., 2012; Liu et al., 2022), show comparable activity in the vps4aT248I mutant with wild-type neurons. This data indicates that perturbation of Vps4a activity does not result in global defects in vestibular function or in brain function in general. Instead, specific circuits are more sensitive to mutations in vps4a, likely due to a greater dependence on processes requiring membrane scission such as high turnover of plasma membrane proteins or intercellular exosomal signaling. Normal VIEMs and activation of TAN and SVN neurons also suggests that excitation of utricular hair cells and vestibular afferent neurons is unaffected in vps4aT248I mutants. In contrast to the vestibuloocular pathway, the vestibulospinal circuit is impaired when vps4a is mutated. VS neurons, which presumably activate trunk motor neurons, have decreased neuronal activity in mutants. Although we used nonstartle tones and selected regions anterior to the Mauthner cell dendrite for quantification of the VS neuronal activity, we cannot exclude the possibility of a small contribution from Mauthner cell dendrites to the overall signal, and thus, our analysis of the VS region likely underestimates the reduction in VS activity. Indeed, analysis of single VS neurons identified by spinal backfills revealed a stronger deficit. Recently it was shown that ablation of VS neurons leads to modest defects in posture in zebrafish larvae (Hamling et al., 2023a), suggesting that other vestibular circuits contribute to postural control. As such, the reduced activation of the vestibulospinal pathway in vps4aT248I mutants may only partially account for the inability to maintain an upright posture.
Further evidence for a loss in central control of motor responses emerged from our analysis of the absence of an acoustic startle reflex in vps4aT248I mutants. Despite the lack of response to aversive acoustic stimuli, vps4aT248I mutants have comparable activation of the ascending auditory pathway as seen in wild-type siblings, with the exception of auditory afferent neurons, which show a partial reduction in function. This partial loss of activity, however, is unlikely to fully account for the strong deficit in behavior. Global defects in the descending motor pathways or the spinal cord are also not the cause for the absence of the motor output in response to visual or tone stimulation. vps4aT248I mutants can exhibit spontaneous bouts of swimming and aversive touch stimuli activate escape responses, demonstrating that mutant larvae can generate large C-bend tail movements. The above experiments indicate that saccular afferent neurons are functional, albeit at reduced levels using nonstartle tones, and Mauthner cells are robustly activated by touch in vps4aT248I mutants, implying that a second pathway may be defective in generating acoustic startle reflexes. One such pathway may involve hindbrain spiral fiber neurons, which have been shown to be essential for Mauthner cell-dependent escape responses (Lacoste et al., 2015). Nevertheless, our in vivo imaging experiments using a nonstartle tone stimulus revealed a pronounced reduction in the activity of midbrain nMLF neurons and downstream spinal cord neurons of vps4aT248I mutants. Neurons of the nMLF play a key role in receiving sensory input and sending motor commands to the spinal cord and are thought to play an analogous role to the interstitial nucleus of Cajal in the mammalian midbrain, which controls eye and head movements (Fukushima, 1987; Severi et al., 2014; Thiele et al., 2014). In zebrafish, both visual and vestibular systems provide sensory input to the nMLF (Bianco et al., 2012; Matsuda and Kubo, 2021; Liu and Bagnall, 2023). Interestingly, TAN neurons of the vestibuloocular pathway also play a dual role of controlling eye and body movements in zebrafish and were recently shown to transmit signals about the roll axis of the body to the nMLF, implicating this circuit in fine control of posture (Sugioka et al., 2023). In vps4aT248I mutants, TAN neurons are also robustly activated by the tone stimulus, yet postural defects are pronounced, further bolstering the notion that sensory input such as vestibular cues fail to trigger sensorimotor transformation within the nMLF.
In summary, our combined biochemical and in vivo analyses have revealed that loss of Vps4a function generates sensorimotor processing defects in the CNS of zebrafish larvae. Our study provides a better understanding of the cause of motor dysfunction when vps4a is mutated. Recent studies of exosome therapy for sensorimotor recovery after injury or resulting from acute or chronic neurodegenerative disease (Chen et al., 2020; Yari et al., 2022; X. Zhang et al., 2023) suggest that exosome therapy may also ameliorate the symptoms of neurodevelopmental disorders such as those associated with VPS4A.
Footnotes
We thank the past members of the Genetics Department of the Max Planck Institute in Tuebingen, Germany, particularly Tatjana Piotrowski and Stefan Schulte-Merker for their assistance in a small-scale allele screen. We also thank Matthew Esqueda and Sivan Brodo-Abo for their help with animal care. This work was supported in part by a National Institute on Deafness and Other Communication Disorders Award (DC0170046) and SICHL funding from the OHNS department of Stanford University to T.N.; a grant from the National Natural Science Foundation of China to S.S. (92254306); and Swiss National Science Foundation (310030_204648) to S.N.
The authors declare no competing financial interests.
E.S.’s present address: Biotechne Corporation, Newark, California 94304. M.H.’s present address: Department of Otolaryngology, University of Colorado, Aurora, Colorado.
- Correspondence should be addressed to Teresa Nicolson at tnicolso{at}stanford.edu.
